WORKING
PAPER
ALFRED
P.SLOAN
SCHOOL
OF
MANAGEMENT
fir
Tfc]^
I
JUN
101989
June, 1989
THE
DYNAMICS
OFR5D COMMUNITIES:
IMPLICATIONS
FORTECHNOLOGY STRATEGY
by
Michael
A.Rappa
WP^3023-89-BPS
MASSACHUSETTS
INSTITUTE
OF
TECHNOLOGY
50
MEMORIAL
DRIVE
CAMBRIDGE,
MASSACHUSETTS
02139
THE
DYNAMICS
OFR^D COMMUNITIES:
IMPLICATIONS
FORTECHNOLOGY STRATEGY
by
Michael
A.Rappa
THE DYNAMICS
OF
R&D
COMMUNITIES:
IMPLICATIONS FOR
TECHNOLOGY
STRATEGY
by
MichaelA.
Rappa
Assistant Professor of
Management
M.I.T.
Presented atthe
CORSATIMS/ORSA
ConferenceMay
8, 1989, Vancouver,Canada
Massachusetts InstituteofTechnology
Alfred P. Sloan School of
Management
50
Memorial
Drive,E52-538
Cambridge, Massachusetts02139
USA
INTRODUCTION
A
central taskofthe research laboratorymanager
istodetermine theoptimalallocation of scarce resources
among
a variety of technologies that could bedeveloped by theresearch staff. It is adifficultand unrelenting challenge with
no
clearanswers and with the optionschanging overtime.
Whether
itis a promisingnew
technology on the horizon or a technology currently indevelopment
that isproving less promisingthan initially thought, the laboratory's ix)rtfolioofprojects
issubject tofrequentreview and reconsideration.
In
what
directions should a research laboratoryexpend
its effon?What
new
technologiesshould be vigorously pursued, andwhat
existing projects shouldbe curtailed? In sorting through these questions, the laboratory
manager must
assess each technology's potential impact on current business, itsrisks, its return,
and estimate the length of time it might take toreach the marketplace—all with an
eye toward
what
might bedone
by competitors.The
time frame orwindow
for a technology isparticularlycritical to the assessment.Even
though the potential ofa technologymay
seem
significant, its importance will increase or diminish dependingupon
the lengthof time itwill take todevelop.There
isno
easy formula forestimating a technology'swindow.
Over
the past several decades theeffort todevelop the field of technological forecasting hasyielded a limited
number
ofapproaches, but even somost
firms continue to relyheavilyon
expen
judgement.'The
benefits and limitationsofexpen judgement
arefairly well understood: in short, experts in a given technology are the
most
knowledgeable tojudge it, butthey are
more
likely tooverestimate itspotential and'Sec for example. Fusfeld and Spilal. 'Technology Foreca5ling and Planning in the Coqxsraie Environment," 1980, and Mariin and Irvine, Foresighi in Science, 1985.
underestimatethe degree ofdifficulty in bringing itto fruition.2 Moreover, itisnot
unusual to find that foreveryoptimistic opinion an equally pessimisticone can be
found.
Given
that resources are limited, thedetermination over the worthiness ofdeveloping a particular technology
may
place a laboratory's researchers at odds withone
anotherand the resulting debate can reach an impasse. This canmake
laboratory life interesting forone
who
enjoys hearty exchanges, but itisno
solacefor the
manager
who
needs to take action andmake
effective allocation decisions.Indeed, the entire laboratory atmosphere can
become
strained,when
researchersbecome
impatient with the slowness in approvingnew
projects andmanagers
become
impatient waiting for investments inongoing
projects to yield tangibleproducts or processes.
The
discovery of superconductivity at high temperatures in bulk ceramicmaterials (namely, the
compound
of lanthanum-barium-copper-oxide) in1986
servesas a excellent
example
ofthechallenge posed by anemerging
technology.3
The
event,which
occurred at theIBM
research laboratory in Zurich, Switzerlandby two
scientistswho
laterwere
awarded
aNobel
FYize for their effort, isconsidered today to be extremely significant in terms of both its scientific and
technological implications and indeed,
some
believeon
thesame
scale as thediscovery of the transistor effect in semiconductor material at Bell Laboratories
forty yearsago. Likethe transistor,itcould ultimately lead tovastimprovements in
areas such as high-speedcomputing.
However,
the realization ofacomputer
withcomponents
based upon thenew
superconducting material is not atrivial task norisitcertain whetherit could be achieved-let alone when. Several problems will have
to be addressed, such as, refining the crystalline structure of the material,
improvingits electrical characteristics, fabricatingitintouseful devicesand circuits
in high volume, packaging thecomponents, integrating these
components
into theotherpartsofthe system, and resolving the scientific question of
why
the materialsbehave astheydo.
^See Anderson,Long-Pange Forecasting, Chapter 6,for adiscussionofjudgmental methods. Also see Luukkonen-Gronow, "Scientific Research Evaluation," 1987, forareview ofqualitativeand
quantitative methods.
^See Muller and Bednorz,"The DiscoveryofHigh-TemperarureSuperconductors," 1987, and Foner andOrlando, "Superconductors; The Long Road Ahead," 1988.
The
anticipated speed inovercoming
theobstaclesfacing theapplication ofsuperconducting ceramics can
make
all the difference in deciding the proper allocation ofa laboratory'sresourcesovertime. Yetjudgments
aboutthe probabletime frame for the technology's
development
arevague
at best and opinions areoften divided. Initially, rapid progress leading to
even
more
importantsuperconducting ceramic
compounds
(yttrium-barium-copper-oxide, inparticular) generatedwidespread
enthusiasm
for nearterm
commercialization of thetechnology.
However,
the reality ofwhat
lies aheadnow
has givenrise to amore
sober opinion
among
some
researchers about the long term nature ofthe effort.TTie perils of this situation are readily apparent to the laboratory manager: ifone
accepts the opinion that such a
computer
can be realized within five years, theappropriate allocationof resources willbe substantially different than ifone holds
the opinion that such a
computer
can be realized only within fifteen years. Ironically, itwas
thesame
firm,IBM,
which
beginning in the early 1970'sattempted todevelop a superconducting
computer
(usingniobium
alloys), but had to scale back its effort in 1983 after reportedly spending asmuch
as one-hundredmillion dollars withoutsuccess.'^
The
case of superconducting ceramics is not unique. In thepast,managers
have wrestled with similardecisions and they will continueto
do
so in the future.5
Time
and again, theymust
grapple with the laboratory'sresearch agenda, seekingto understand
what
new
technologies are gainingmomentum
andwhat
ones aregrinding to a halt at the researcher's bench.
The
purpose of this study is to assist the research laboratorymanager
inunderstanding the rate of progress in the
development
of technology in order toimprove
their effort to optimize resource allocation. Specifically, a theoreticalmodel
isproposed forassessing the developmentalrate of anemerging technology.The
model
focuses on thecommunity
of researchers that coalesces around a^See Robinson. "IBM Drops Superconducting Computer Project," 1983.
'Recent claims regarding "cold fusion," ifverified, may hold asimilar challenge as thatof high-temperature superconductors. See Pool, "Fusion Breakthrough?" 1989.
technology: that is, the scientists and engineers
who
arecommitted
to solving aninterrelated set of scientific
and
technical problems,and
who may
beorganizationally and geographically dispersed, but
who
neverthelesscommunicate
witheachother.
The model
seeksto uncoverthe relationship betweenthe structuraland behavioral
dynamics
of this"R&D
community"
and its rate ofprogress insolving the
mjtiad
ofproblems it faces.The
theory supports the contention thatcertain changes in the structural and behavioral characteristics ofthe
community
may
berelated tothe acceleration or decelerationofatechnology'sprogress toward commercial introduction.R&D
COVnv^UNTTIESIn his influential
work The
Limits of Organization,Kenneth
Arrow
statesthat "...organizations are a
means
of achieving the benefits ofcollective action insituations
where
inwhich
the price system fails." Following in a similar vein, itmay
be thatR&D
communities
are ameans
ofcollective action in situations inwhich
the firm (or managerial hierarchy) fails.Given
thisperspective, itmay
behelpful to
view
suchcommunities
as efficient meta-organizational structuresfor accomplishing certain goals reasonablybeyond
the reach of the individualresearcheror team.
Despite the fact that the research
community
is a familiar concept in thecontextofthe scientificworld, itsplace in therealmoftechnological development is
largely
ambiguous.
^ It is well understood that scientists (particularly thoseemployed
in universityorgovernment) aremembers
ofcommunities, the so-called invisible colleges, inwhich
information flows with relativefreedom between
laboratories.
These communities
provide themechanism
bywhich
members
meteout recognition and rewards and set the direction for future research. In contrast,
technological
development
is typically seen as thedomain
ofengineers and the*Thehistorian EdwardConstant, whose study ofthe developmentofthe turbojet elucidates the role ofthe technological community, stales: "While extensive research has been done on 'invisible colleges,'research fronts, and the community structureofscience, there has been little analogues sociological or historical investigation of technological practice." (Constant, 1980, p. 8.) For earlier studies of scientific communities see Griffilh and Mullins (1972) and Mullins (1972).
industrial firms that
employ
them. Firms operate to establish proprietaryknow-how,
which
then can be leveraged to developnew
products or processes thatsurpass thoseofcompetitors. Secrecy, competition, and managerial direction are
the sine
qua non
of the technological landscape.Given
this traditionalconceptualization of technological development, it appears that the notion of a
community
of researchersisatonce incongruentA
closer examination of science and technology yields exceptions to suchbroad stereotypes.
To
view
scientificcommunities
as friendly clubs inwhich
members
freely share their ideas is misleading.Community
members
are notimmune
to fierce competition, racing to stake intellectual claims (typically in theform
ofjournal articles, but increasingly in theform
ofpatents'') and,even thoughit
may
be contrary to established scientific norms, acting to restrict the flow ofinformation abouttheirresearch.
Likewise, the world oftechnology is equallyas complex. Firms compete,
but they also cooperate with each other, allowing technical information to flow
among
engineersin different organizations.8Some
engineers attend conferences,present technical papers, and publish the results of their
work
in peer-reviewjournals sponsored by professional societies. Like scientists, they too,
may
seethemselves as
members
ofa particularR&D
community,
which extendsbeyond
the boundary oftheir firm. Indeed,some
are scientists, in that they are trained in thescientific
method
andmay
havedoctoral degrees.To
the extentthat researchers inaparticular technological
domain
consider themselvesmembers
of aR&D
community,
thiscommunity
may
play an instrumental role in influencing the rateand direction of thetechnology's development. Contrary toestablished opinion, the
development
of anew
technologymay
not simply the activity of a handful of''Although ii is notnew, majorresearchuniversities, such asMITand Stanford, arepaying
increased aiteniion to opportunities to patent the inventions of faculty. MIT has a well-organized staff of professionals dedicated topatenting and licensing activities. See Eberlein (1989) for a description of
technology patenting and licensing activities atMIT. The importance of patenting is illustrated in the case of cold fusion research withMIT's recent announcement(viapress conference) that it applied for patents ona theoreticalexplanation of cold fusion, justone day after the theory was formulated (see
"Fusion in a Bottle," 1989).
'The flowof informationamong firmsisdocumented in the workof Allen (1979),andrecently ha.s received attention from von Hippcl (1987) in his examination ofknow-how trading.
engineers, or of a firm, but instead
many
individualsworking
innumerous
organizations spread throughouttheworld.9
In the context of this study, the
R&D
community
is defined to includeindividuals in any type oforganization, such as universities, private firms,
new
ventures, quasi-public corporations, and
government
research institutes, andfurthermore, the
community
can be global in scope. This concept of aR&D
community
is obviouslymore
inclusive than thatof an industry,but evenmore
soin that it includes firms with
no
regard towhat
industry theymay
be typicallyclassified (forexample, semiconductors, computers, chemicals, or aerospace) orno
matter what segmentthey
may
operate in (such aswith thesemiconductor industry,there are
segments
dedicated to materials, devices, process equipment,and
soforth), so long as they are focusing
on
some
part ofthe relevant set ofproblems.Thus, the
community
is definedby
theproblem
set and not necessarily by the output, as isnormally thecase with anindustry definition.THEORETICAL CONSIDERATIONS
Inreality, technological
development
is a vastlycomplex
process. But forthis purpose it will be treated as a simplified theoretical abstraction. It is readily admitted that the theory does not adequately describe all of the subtleness of
technological
development
as it might actually occur.However,
the intent is tocapture the essence ofthe process such that a basic understanding of the rate ofa
technology's
development
can be established.The
foundation of themodel
rests on the principle that technologicaldevelopment is an intellectual process
whereby knowledge
iscreated andappliedtoform
new
products orprocesses.The
creation and application ofknowledge
entails'Tlienotion ofa
R&D
community hasbeen obscured by the tendencyto viewtechnologicaldevelopment in teirns of inventorsand their inventions, such as with the Shockley, Bardeen and
Brattain andthe transistor. However, acareful historicalaccount shows "It is...unrealistic to see the transistor as a product ofthree men,or ofonelaboratory,orof Physics, oreven ofthe forties. Rather its invention requiredthe contributions ofhundreds ofscientists, working in many different places, in
resolving certain
problems
which
stand in theway
of realizing the desiredoutcome.
10The
central actors are the individual researcherswho
become
committed
to solving the problems, and it is theywho
set the process in motion.The
basicelementsofthemodel
areoutlined below.(1) Let the
body
ofknowledge
in a givenrealm
of technology berepresented as
K.
Given
the state of the world at the time of the technology'sinitiation, the
body
ofknowledge
is projected as an increasing function over theinterval oft (0,«=) represented as K(t) in Figure 1(a).
Assume
thatK
can bedecomposed
into its constituentpans
represented as k,which
arehomogeneous
units of
knowledge,
where
K
=
Zk.(2)
Given
K(t), letthe level ofproblems to be resolved (P) be an inverselyproportional function decreasing over the interval t {0,«»), represented as P(t) in
Figure 1(b).
Assume
thatP
also can bedecomposed
into its constituent parts,which
arehomogeneous
units of problems,where
P
= Zp.
The
relationshipbetween
the state ofknowledge
and the level ofproblems to be solved is given asP(t)
=
u/K(t),where
u isconstant.At
time (t=
1), given the assumptionsabout K(t)and P(t), a projection canbe
made
based onexpectations about the future point in time of commercialization(tc), where the level of
knowledge
anained (Kc) is sufficient toreduce theproblems toa level (Pc) necessary to apply the technology to practice. Notice that it is notassumed
allproblems
will be resolved, and, indeed, theproblem
solving activity continues after the technology's initialmarket
entry. Furthermore, P(t) isasymptotic to thex- and y-axis, implyingthat atthe initial stage (t --> 0),
P
is not ascertainable, and that as even as time progresses into the future (t —>°°),some
problems
will never be satisfactorily resolved—it is the latterwhich
gives rise tonew
technologies. It is also important to note thateven
given the initialassumptions about K(t) and P(t), the probability of finding a solution to each
'OThe idea ihal technology is essentially knowledge has gained acceptance among scholars in
several disciplines. For example, see Arrow (1962), Layion (1974), Constant (1980), and Aitken
(1985). The work ofKalz and Allen (1985) focuseson technology development as aninformation processing and problem-solving activity. The intenthere is to link the concepts of information,
8
problem
is uncertain.n
Lastly, Ps (=
P, - Pc) over the interval (t>
c), will bereferred to as the
problem
set: a relatively large value of Ps implies a technologywhich
isa radical departurefrom
the established base ofknowledge
toreach tc. Incontrast,a relativelysmall value of Pj implies atechnology
which
is incrementalinnature; that is, it is
more
an extension than a departurefrom
existing base of knowledge.(3) Let Ri (
= Xr)
represent thecommunity
of researcherswho
attempttocreateand apply
knowledge
in order toresolvetheproblemsfacing thetechnologyattime t.
Assume
they arerational, in theeconomic
sense thatthey aremotivatedby
self-interest.The
researcherconductstwo
basic activities: (a) the productionand
communication
of information, and (b) thetransformation of information into theknowledge
required to solve problems. Information production implies thecreation of
new
data through observation and experiment.Communication
ofinformation implies that the researcher can also gather information produced by
another researcher,
and
disseminate to others the information he produces.Knowledge
is a distinct entityfrom
information in thatknowledge
enables the researcher todo something
(know-how)
or explainsomething
(know-why).
Having
information does not necessarily imply either; the researcher firstmust
make
sense oftheinformation availabletohim
inorderto solve problems.12The
information production,communication
and transformation processesare time
consuming;
therefore, any individual researcher can only accomplish a certain level ofeffortin agiven period oftime.Assume
that ina single period oftime, the
amount
ofinformation generated bya single researcher (r) is i, and thathe can perform
m
transformations.The
information he produces andhow
hechooses
to transform it intoknowledge
is an expression of the individualresearcher's
own
creativity, although hemay
be influencedby
the people with"Theconcept of uncertainty in technology development is developed by many authors, including Arrow(1962)who states "Producers [ofknowledge] havetomake decisionsoninputs at the piresent moment, butoutputs are notcompletelypredictable from the inputs."(p. 610) See also, Sahal (1983) who contends "We find...that technological fwogress is a cumulative process of learning to learn
andunlearn in aprobabilistic manner."(p. 216)
'^Theliterature on technological developmenttends totreat informationandknowledge as indistinguishable concepts. See, for example. Arrow (1962).
whom
he collaborates.Not
aJl transformations aresuccessful in yielding a unitofuseful
knowledge.
Note
also,any
given k might be obtainedfrom
differenttransformations, or
from
thesame
transformationimplemented
simultaneously,albeit independently, by different researchers.
The
probability that a particulartransformation will yield a unit of knowledge,
p(m)
=
k, cannot be determined apriori, but rather is learned over time as researchers generate information and performtransformations. Moreover, the relevanceofany piece of information (in
that itmightform partofthe transformation which yields k) cannot be determined a
priori.
The
rate of growth inknowledge
(3K/3t) is a function of thenumber
ofdiverse of transformations
(Zm)
that are performed, subject to the availableinformation
(Si)and
theamount
of additionalknowledge
required for commercialization(Kz =
Kg
- K,): that is,9K/3t
=
/(Xm
I Zi, K^). It isexpected that greater levelsin the diversityof
m
will be associatedwith higherratesin the production of
new
unitsofknowledge. Moreover, the likelihood of successof any given transformation improves as the
amount
of information itdraws
onincreases and as the
amount
ofknowledge
necessary for commercializationdecreases.13
The
researchcommunity,
R, is said to be perfectly contiguous ifforeach r,Xi]
= Xi2 =
Sis •••Zin; that is, information is symmetric. Conversely,R
isperfectly non-contiguous if for each r, Zii '^ Ziz '' Zis-.-Zini that is,
information is asymmetric.
The
degree of contiguity ofR
is influenced by theexistence of organizational boundaries
between
researchers and theireconomic
motivations. Organizational boundaries areimportant for
two
reasons: first theygive rise toinformation asymmetries becausethey
impede
the flow ofinformationand increase the cost of information gathering.
As
a result, organizationalboundaries can slow the rate ofproduction ofk within a
community
by reducing theamount
of information available to each researcher.However,
organizationalboundaries can also enhance therate ofk production to the extent it increases the
"This notion canbe understood interms ofKuhn's (1970) jig-saw puzzle analogy ofscience.
At the start, thepuzzle isquitedifficult,but as pieces are linked togetherand the picturebecomesmore
10
diversity of
m
pursuedby
R
as a whole, since researchers in differentorganizations are likely to
have
different information sets and use differenttransformations.
The
communication impedance
effect of organizational boundaries isovercome
viaresearcherswho
act as technological gatekeepers--thatis,researcherswho
tend tocommunicate
with others in different organizations.14Given
the economically rational behavior ofresearchers, thecommunication
ofinformationacross boundaries occurs as a
form
of quid proquo.n
Boundary-spanning
communication
is defined here ina restricted sense as person-to-p)erson exchanges and does not include publications. (Papers are very limitedmechanisms
forinformation exchange becauseofthedelay in publication, theproblems in encoding
complex
information, and the inability toensure fairexchanges. Published paperscan serve as a
mechanism,
like patents,to stakeclaimstoknowledge.)(4)
The
researcher's objective is tomaximize
thenumber
of units of k heproduces and can lay claim to before otherresearchers.
Each knowledge
unit haspotential value, butit is
assumed
that the ultimate value ofthe researcher's claimsdepends
upon whether
or not allproblems
necessary for reaching the point of commercialization are resolved and the length of time taken. That is,knowledge
has value only
when
it is used, and the soonerit is used the greater value it willhave.
The
researcherdoes not need toproduce all theknowledge
required toreachto it is inconsequential to
him
who
solves theotherproblems, aslongashisclaimstoone or
more
k areestablished.The
researcher's intellectual capital consists oftwo
components: generaldisciplinary
knowledge
that is held incommon
with other researchers, andspecialized
knowledge
that flowsfrom
hiswork
related to a particulartechnology(IM). It is
assumed
that the researcher can contribute to alternative technologieswithin the realmofhisdisciplinary knowledge, and that he is free to switch atany
time.
1 1
Since hischoice isguided by theobjective to
maximize Zk,
theresearcheris motivated to enter only those fields in
which
he believes the transformationprocess has a high probability of yielding
him
units ofknowledge
he can claim.The
value ofp(m)
=
k is learned over time as thenumber
of transformationsresearchers perform increases.
Thus
a researchermay
be encouraged to enter afield he believes might have a high value of p(m), but the actual probability can
only be determined with experience. If
p(m)
is equal to orgreater than expected(i.e., the task of acquiring k is easier than anticipated), then the researcher will remain in the field; however, if/7(m) is lowerthan expected (i.e., acquiring k is
harder than anticipated), then the researcher might switch to another field with
higherperceivedp{m).
The
switching behavior ofaresearcherismoderated by thelength of time he spends contributing to a technology's
development
and theamount
ofknowledge
claims he accumulates.A
researcher's specializedknowledge
(Zk)becomes
a sunk cost: themore
k theresearcheraccumulatesover time, thelesshkely he isto exita fieldpriortotc.(5)
At
this point, it is possible todraw
together the basicelements outlinedabove
into adynamic model
of the process of technologicaldevelopment
(seeFigure 2).i5
To
begin, suppose that at the initiation of anew
technology (t=
1),there are a certain
number
of researchers involved (Ri)who
in each instantoftimeproduce a cenain
amount
ofinformation (Z'l) and perform a certainnumber
oftransformations
(Xm).
The
researchers' effort yields a rate ofgrowth
inknowledge
(3K/3t>
0) which, given theproblem
set confronting them, implies arate of progress toward commercial introduction ofthe technology (5tm/3t,
where
tm
=
tc -11)-
The
resulting rate ofprogress, in turn, influences the decision of aresearcher to enter, remain in or leave the
community.
Since the size of theresearch
community
determineshow much
information isproduced andhow many
transformations areperformed, theentryorexitdecision ofresearchers isofcritical
importance tothe acceleration or deceleration in the productionofknowledge.
Itis important tonotice that the researcher'sdecision toenterorexit a field
'^To be moreprecise, it may be possibletoformulate asystemsdynamic modelof
R&D
communities. Although not presented here, oneis currentlyunder consideration. See Roberts (1981)
12
is not a one-time choice, but rather is subject to consideration over time.
The
attractiveness ofdeveloping a particular technology changes over time, as
more
information is produced andresearcherslearn the probabilities in transforming this
information intoknowledge. Its attractiveness also
may
beinfluencedby
changesin other technologies, and indeed a variety ofother events, all of
which
will bereflected inthe researcher's decision.
Suppose, as time progresses, the degree of difficulty in transforming
information into therequired
knowledge becomes more
apparent and it is lowerthan expected. This will result in an acceleration in therate of
knowledge
growth(32K/9t2
>
0) that implies a point ofcommercialization sooner than anticipated.Some
researchers will respondtothischange inconditions byentering into thefield(3R/8t
>
0),which
will contribute to the production of information andtransformation process and feed the acceleration further. Conversely, suppose the
transformation process performed
by
researchers is provingmore
difficult thanexpected. This will result in adeceleration in the growth of
knowledge
(32K/3t2<
0) and,consequently, thepointof commercialization will be shifted furtherout into the future. This is a negative signal to potential entrants and
may
also lead to theexit of researchers
from
the field (subject to their sunk cost considerationsmentioned above.)
The
effectsofthese changes areillustrated in Figure 3 and4.The
impact oftheflow of researchersinto a field ondKJdt
ismoderated bythe burgeoning
community's
structure and behavior: that is, the contiguityof thecommunity
as determined by the distribution ofresearchers across organizationsand the
communication
between them.To
illustrate thispoint,assume
theextremeconditionsofperfectcontiguity and perfect non-contiguity. In the firstcase, as the
research
community
grows, all researchers areemployed by
thesame
organization.Thus
all researchers areassumed
tohave thesame
information set, but the diversityoftransformationsperformed is limited by the researchers'mutual influence. Inthe
second case, as the
community
grows, each researcher isemployed
in a separate organization. Thus,each researcheris working from a different set ofinformationand
performing
independent
transformations, such that the diversity oftransformations is maximized.
Simply
stated, in one situation each researcher in13
transforming that information into knowledge; in theother situation, althoughthe
community
generates thesame amount
information, theamount
available to any particularresearcheris small and the varietyof approaches takenis greatSTRATEGIC
CONSIDERATIONS
Technology
strategy has tended to focuson
the firm and industry as theprincipal units of analysis. It is proposed here that the concept of an
R&D
community
may
have something to add to our understanding of technologicaldevelopment,
and thusmay
prove
useful in the strategic considerations of managers. This section will review the possible benefitsof anR&D
community
perspective totechnology strategy.
First, examination of the
R&D
community
broadens attentionfrom
theanalysisof firms in a panicular industry (as define by its products), toall typesof
organizations in various types ofindustries and sectorsofthe
economy.
Thismore
comprehensive perspectivecan bemost critical in understanding the
emergence
ofanew
technology, since awhole
new
industrymay
form from
the efforts of thesedisparate organizations.
For
example,
in the case of theemergence
ofsemiconductor technology, it lead tothe formation ofa
new
industry distinctfrom
the industry
based
on
the establishedtechnology
(that is,vacuum
tubemanufacturers).16
Second, analysis of
R&D
communities
may
provide for betteranticipation ofemerging
technologies. It is typical foremerging
technologies to be indevelopment
for a decade ormore
prior to commercialization. Therefore, theevolution of the
R&D
community
can precede the formation of an identifiableindustry
by
many
years. Strict focuson
the industrymay
lead to chroniclags in afirm's technology developmentefforts.
Third, focusingon the
R&D
community
may
providemanagers
withinsight into the structure of the social networkamong
researchers andcommunication
14
flows.
The
evolvingnetwork structuremay
be an early signal to the formation of strategic alliancesbetween
firms as the technology approachescommercial
introduction. Also, analysis of the network structure will enable the firm to
determine whetheror not it is adequately--as researchers say--"plugged-in to the
grapevine."
Where
afum
sitsin the network determineshow
much
infoimation isavailable to its researchers. TTie firm does not have to be a central
node
in thenetwork, so long asit
knows
and interacts with those organizationswhich
are thecentral nodes.
Fourth, the
R&D
community
perspective highlights the collective actiondimension of technological development, and in this
way
it addressesdirectly thestrategic interdependence
which
may
arise. Unless a researcher or a firm canproduce
all the necessary information and resultantknowledge
required forcommercialization ofatechnology, thenthe technology's developmentis ultimately
a collective action
problem
where
the decisions of researchers and firms areinterdependent.
Managers
can view themselves in a adversarial position,where
theobjectis tomonopolizeclaims toall units of
knowledge
withrespecttoapaniculartechnology; or
mangers
can seethemselves faced with a collective action problem,in
which
the goal is toestablishsome
claims toknowledge
while at thesame
timepromoting the technology's development
among
other organizations sothat all thenecessary
knowledge
will be established.The
approach takenhas implications forthereturn, rate of development, andrisk facing the firm. In the first case, the return is relatively high, but the speedat
which
the technologydevelopsmay
be slowerandthe riskofabsolute failure intheeffort great. In comparison, in the second case the return
may
be lower, but therateofdevelopment fasterand theriskofabsolutefailureless.
The
well-known
historicalexample
that illustrates this concept is thedevelopment
ofsemiconductor technology beginning with thetransistor. Aftertheinvention of the point contact transistor at Bell Laboratories,
AT&T
pursued astrategy to
promote development
of the technology broadly holding symposia totransfer information and
knowledge
about the transistor to all other interested organizations, and offering low-royalty Hcenseson the transistorpatent.The
effect15
on the growth ofa semiconductor
R&D
community was
noticeable, asmany
more
organizations did
become
involved in developing the transistor. Indeed,many
critical advancementsin semiconductortechnology necessaryforcommercialization
ofthe technology
came
from
researchers outside ofBell Laboratories, researcherswho
might not have taken part in the technology'sdevelopment
had it not been forAT&Ts
promotional efforts.CONCLUSION
This paper proposes the idea that the
"R&D
community"
is a usefulconceptin understanding the rateofatechnology's development toward commercialization.
An
initial theoretical exposition is provided, and a brief discussion of theimplications for technology strategy. Clearly
much
more
research is required tofully understand theimportanceofthis conception oftechnological development.
A
researchprogram
focusedon
the study ofR&D
communities
intechnological
development
iscurrently in progress. This research seeksto identifyand
measure
changes over time in the structural and behavioral characteristics ofR&D
communities
and the relationshipbetween
these factors and the rate ofprogress achieved. Technologies underinvestigation include:
-GALLIUM
ARSENIDE INTEGRATED
dRCUTTS
-JOSEPHSON
JUNCTION
DEVICES
-REDUCED INSTRUCTION
SET
COMPUTERS
-IvTEURAL
NETWORK
COMPUTERS
-MAGNETIC
BUBBLE
MEMORY
STORAGE
DEVICES
-HIGH-TEMPERATURE
SUPERCONDUCTING
MATERIALS
-POLYPROPELENE PROCESS
TECHNOLOGY
-EPDM
RUBBER
PROCESS
TECHNOLOGY
-MICROGRAVITY MATERIALS PROCESSING
Although
these studies are already yielding a wealth of information aboutR&D
communities, furtherresearch is required for insight intowhy
R&D
communities
16
function as they
do
and toprovide better a understanding of the implications for17
BreLIOGRAPHY
Allen,TJ.,Managing theFlowof Technology (Cambridge:
MIT
Press, 1977).Annstrong,J.S.,Long-RangeForecasting,2ndEdition(New York:Wiley, 1985).
Arrow,K.J., "Economic Welfare andihe Allocation ofResources forInvention" in The Raleand Direction ofInventiveActivity,
NBER
(Princeton: PrincetonUniversityPress, 1962).Arrow, Kenneth J.,TheLimitsofOrganization (New York: Norton, 1962).
Braun,E.andMacdonald,S., RevolutioninMiniature:The HistoryandImpact of Semiconductor
Electronics,2ndEdition(Cambridge,CambridgeUniversityPress, 1982).
Constant, E.W., The Origins ofthe Turbojet Revolution (Baltimore: Johns Hopkins University
Press, 1980).
Dasgupta,P. "Patents, Priority and Imitation or.The Economics ofRaces and Waiting Games," TheEconomicJournal 98 (1988): 66-80.
Dasgupta, P. and David, P. A., "Information Disclosure and the Economics of Science and Technology," in G. Feiwel (ed.). Arrow and the Ascent ofModern Economic Theory (New York:
NYU
Press, 1987).Eberlein,J.A., "TechnologyTransferat MIT:
An
Analysis oftheTechnology LicensingOffice,"S.M.Thesis,
MIT
(Cambridge,Massachusetts) 1989.Foner, S. and Orlando, T.P., "Suf)erconductors: The Long Road Ahead," Technology Review,
(1988): 36-47.
Fusfeld, A.R. and Spital, F.C., "Technology Forecasting and Planning in the Corporate
Environment," in TIMSStudies in theManagementSciences, 15,(1980): 151-162.
"Fusionin a Bottle: Miracle orMistake?"BusinessWeek,(May 8, 1989): 100-110.
Griffith, B.C. and Mullins, N.C., "Coherent Social Groups in Scientific Change," Science, 177
(1972): 959-964.
Katz, R. and Allen, T.J., "Organizational Issues in the Introduction of
New
Technologies," inP.R. Kleindorfer (ed.). The Management of Productivity and Technology in
Manufacturing (PlenumPress, 1985).
Kuhn, T.S. The Structure of TechnologicalRevolutions, 2nd Edition (Chicago: University of
Chicago Press, 1970).
Layton,E.T., "TechnologyasKnowledge," TechnologyandCulture 15 (1974): 31-41.
18
Mijllcr,K.A. and Bcdnorz,J.G.,"The Discovery ofHigh-Temperature Superconductors," Science,
237(1987;: 1133-39.
Mullins, N.C., "The Developmentofa Scientific Specialty,"Minerva 10 (1972): 51-82.
Pool.R.,"Fusion Breakthrough?" Science243 (1989): 1661-1662.
Roberts, E.B.,ManagerialApplicationsandSystemDynanvcs(Cambridge:
MIT
Press, 1981).Robinson, A.L., "IBM Drops Superconducting Computer Project," Science. Ill (1983): 492-494.
Rogers, E.M., "Information Exchange and Technological Innovation," in D. Sahal, (ed.), The
TransferandViilizalionofTechnicalKnowledge(Lexington: Heath, 1982).
Sahal, D. "Invention, Innovation and EconomicEvolution," TechnologicalForecastingandSocial
C/w/igf23 (1983): 213-235.
Sahal, D. "Technological Progress and Policy," in D. Sahal (ed.),The Transferand Utilizationof
TechnicalKnowledge(Lexington: Heath, 1982).
von Hjpf)el,E.A., "Cooperation BetweenRivals: Informal
Know-How
Trading," ResearchPolicyK
Kt PcFIGURE
1 (a) icFIGURE
2KNONM-EDGE
INFORMATION
R.ATEOF
PROGRESS
LNDEV'ELOPMENT
RESEARCH
COMMUNIPJ'
nCfRE
3 nGLTRF.4 Kt Pc Kt Pc IC t tcDate
Due
NOV
13
US
JUN
119:}0
MIT LIBRARIES DUPl 1
3 TDflD